Two major clinical problems in patients with malignancies include: (1) resistance of their tumors to antineoplastic agents and (2) continued progression of their tumors (local invasion and metastases). The mechanism(s) by which these problems arise are incompletely understood. The program outlined proposes to develop a model which can be used to study both of these clinical problems. The proposed model is based on strong preliminary evidence that one way tumors develop, progress and become resistant to antineoplastic agents is through a unique mechanism for diversification of their genome. This mechanism involves excision of pieces of genomic DNA which may contain oncogenes or drug resistance genes and an origin of DNA replication. These circular molecules (episomes), when amplified may contribute to tumor progression (if the amplified gene is an oncogene) or to drug resistance (if the amplified gene is a drug resistance gene). The specific aims of this proposal include: 1) Optimization of the methods for isolating episomes from tumor cell lines and from primary and metastatic human tumors. 2) Determination of methods to eliminate these episomes or other extrachromosomally amplified sequences from tumor cells. 3) Basic studies of the mechanisms by which currently used treatments (chemotherapy) induce episome formation. The team which has been brought together to accomplish these objectives is composed of basic and clinical scientists. We have limited the proposal to the Salk Institute, where the molecular biology expertise is centered (Dr. Wahl) and the University of Texas where there is expertise in clinical research and episome analysis in primary human tumors (Dr. Von Hoff). This proposal is a natural extension of projects which were developed during the sabbatical year spent by Dr. Von Hoff in Dr. Wahl's laboratory. With the development of the models outlined here, a clearer understanding of the mechanisms of tumor progression and drug resistance should emerge and result in new therapeutic approaches which can be rapidly applied to clinical oncology situations.